World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
CFD simulation of Pressure Drops in Liquid
Acquisition Device Channel with Sub-Cooled
Oxygen
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
David J. Chato, John B. McQuillen, Brian J. Motil, David F. Chao, and Nengli Zhang
Abstract— In order to better understand the performance of
screen channel liquid acquisition devices (LADs) in liquid oxygen
(LOX), a computational fluid dynamics (CFD) simulation of LOX
passing through a LAD screen channel was conducted. In the
simulation, the screen is taken as a ‘porous jump’ where the pressure
drop across the screen depends on the incoming velocity and is
formulated by Δp = Av + Bv2. The CFD simulation reveals the
importance of the pressure losses due to the flow entering from
across the screen and impacting and merging with the channel flow
and the vortices in the channel to the cumulative flow resistance. In
fact, both the flow resistance of flows impact and mergence and the
resistance created by vortices are much larger than the friction and
dynamic pressure losses in the channel and are comparable to the
flow resistance across the screen. Therefore, these resistances in the
channel must be considered as part of the evaluation for the LAD
channel performance. For proper operation of a LAD in LOX these
resistances must be less than the bubble point pressure for the screen
channel in LOX. The simulation also presents the pressure and
velocity distributions within the LAD screen channel, expanding the
understanding of the fluid flow characteristics within the channel.
Keywords— Liquid acquisition devices, liquid oxygen, pressure
drop, vortex, bubble point, flow rate limitation.
I. INTRODUCTION
I
N future space vehicles, cryogenic propellants are usually
transported from their storage tanks to an engine in liquid
state. This is easily accomplished on earth using gravity
provided that the tank outlet is placed at the bottom. Under the
effect of earth gravity, buoyancy effectively separates the
gas/vapor and liquid in the tank and prevents the gas/vapor
mixing into the liquid propellants in the transport channels.
However, the acceleration forces and difficult thermal
environment conditions in space complicate the storage and
transport of propellants, especially with regards to separating of
the gas/vapor and liquid. In low gravity or reduced acceleration
David J. Chato is with NASA Glenn Research Center, Cleveland, OH
44135 USA (e-mail: [email protected]).
John B. McQuillen is with NASA Glenn Research Center, Cleveland, OH
44135 USA (e-mail: [email protected]).
Brian J. Motil is with NASA Glenn Research Center, Cleveland, OH
44135 USA (e-mail: [email protected]).
David F. Chao is with NASA Glenn Research Center, Cleveland, OH
44135 USA (e-mail: [email protected]).
Nengli Zhang is with Ohio Aerospace Institute at NASA Glenn Research
Center, Cleveland, OH 44135 USA (corresponding author to provide phone:
216-433-8750; fax: 216-433-8050; e-mail: [email protected]).
International Scholarly and Scientific Research & Innovation 3(10) 2009
environments in space missions, surface tension, rather than
buoyancy, becomes dominant in determining the relative
positions of liquid and vapor propellants during quiescent or
coast periods. During the high acceleration engine thrust
period, single-phase liquid is delivered from the tank simply
by “draining” the liquid from the tank “bottom” and using an
anti-vortex baffle over the tank outlet to minimize gas/vapor
ingestion due to the flow swirl. In other flight periods of space
vehicles, single-phase liquid withdrawal is a challenge in low
gravity because liquid may not cover the tank outlet.
Consequently, the investigation of the performance of surface
tension propellant acquisition/expulsion devices has been a
critical research and development concern for space missions.
Many types of these devices have been built, tested and used
for hypergolic propellants, such as nitrogen tetroxide (N2O4)
and monomethyl hydrazine (MMH), but development for
cryogenic propellants has been lagging. These devices are
designed to achieve a unique set of performance requirements
under appropriate environmental conditions; there is no
universal design that satisfies all mission criteria for all
applications [1]. Future space vehicles require using non-toxic,
cryogenic propellants, not only stemming from the offering of
performance advantages over the toxic hypergolic propellants
but also from environmental and handling concerns [2]. For
cryogenic propellants, the “settling” method previously used
to position propellants over the tank outlet will have to be
relinquished, due to conflicting vehicle requirements. A
capillary flow liquid acquisition device (LAD) is required for
propellant management [3].
Capillary flow LADs have been well characterized for
storable toxic propellants [2], [4], but there are only a few
LAD experiments and data for cryogenic propellants,
especially with liquid oxygen (LOX). Kudlac and Jurns [2]
made the first known non-proprietary effort to measure bubble
point and to collect LAD data in LOX. Jurns and McQuillen
[5] extended the range of LOX fluid conditions examined, by
reporting on bubble point testing with sub-cooled LOX, and
provided insight into factors that may predict LAD bubble
point pressures. Limited by the measuring technologies, the
detailed pressure and velocity distributions in the LAD
channel have been unknown. Computational fluid dynamics
(CFD) simulation can provide this essential information, thus
leading to a better understand LAD screen channel
782
scholar.waset.org/1999.7/15651
World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
performance for LOX.
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
II. CFD MODEL OF LAD CHANNEL
Under NASA’s continuing Cryogenic Fluid Management
(CFM) development program, tests will be conducted in a
prototypical screen channel LAD using LOX at flow rates
representative of those for the main engine for the Lunar Stage
Ascent Module. This screen channel LAD is shown in Fig. 1. A
screen of Dutch Twill mesh 200x1400 is welded along the top
rim of the LAD channel. Surface tension and the small pore
openings in the screen mesh prevent gas/vapor from passing
through the screen while allowing liquid to pass freely. The
plate is positioned above the screen mesh to simulate the tank
wall. This cover plate provides additional protection from
gas/vapor penetration into the LAD channel and is a source of
additional flow resistance. A detail of the screen is shown in
Fig. 2. The CFD simulation model is identical with the test
assembly. GAMBIT™ was used to create the computational
grid mesh and FLUENT™ 6.3.26 performed the simulation.
Cover plate
Flow in
Outlet-tube
Flow out
Top view
δ
Side view
Warp wire
Shute wire
Fig. 2 Dutch Twill mesh screen weaving pattern
The basic geometry of the LAD channel is follows. The flow
channel is 609.6 mm (24 in) long by 50.8 mm (2 in) wide by
25.4 mm (1 in) deep. A lip along the top of the channel
facilitates attachment of the screen section to the channel and
forms a screen window of 482.6 mm (19 in) long by 50.8 mm
(2 in) wide. The outlet-tube has an inner diameter of 23.622
mm (0.93 in). The origin of coordinates was set at the center
of screen top surface, as shown in Fig. 1b. Since the screen is
too thin (the screen thickness δ § 0.15 mm) to create the grid
mesh, the screen is treated as a ‘porous-jump’ whose
parameters determined by the screen characteristics.
Based on the experimental data with liquid nitrogen,
Armour and Cannon [6] developed a general correlation of
screen friction factor which is applicable to the flow through
all types of woven metal screens:
LAD channel
݂ ൌ ܽȀܴ݁ ൅ ܾ
a)
Test LAD channel assembly with cover plate
Screen (Dutch Twill 200x 1400)
y
x
z
where f = Δpε2D/(QδρV2), Re = ρV/(μs2D), a = 8.61 and b =
0.52. Δp is the pressure drop for fluid flow through the screen,
while the parameters Q, ε, D, δ, and s are the tortuosity factor,
the void fraction, the pore diameter, the screen thickness, and
the surface area to unit volume ratio, respectively, while ρ, μ,
and V are the density, viscosity, and incoming velocity of the
fluid passing through the screen, respectively. The parameters,
Q, ε, D, δ, and s were determined by Armour and Canon based
on their analysis and measurements for each specified screen.
In fact, (1) can be directly written as
߂‫ ݌‬ൌ ‫ ܸܣ‬൅ ‫ ܸܤ‬ଶ
Blocked
b)
Test LAD channel assembly without cover plate
Fig. 1 Test LAD channel assembly
International Scholarly and Scientific Research & Innovation 3(10) 2009
(1)
(2)
where A = 8.61Qδμs2/ε 2; B = 0.52Qδρ/ε2D.
Using Armour and Cannon’s data and experimental results
for liquid hydrogen, Cady [7] found that the constants a = 8.61
and b = 0.52 in (1) provide good agreement for the screen of
Plain Dutch mesh 24x110, while considerable errors occur in
other screens. He revised (1) with different values of a and b
for each specified screen. For the screen of 200x1400 Dutch
Twill mesh, a = 4.2 and b = 0.20, while Q = 1.3, ε = 0.248, D
= 0.000010 m, δ = 0.0001524 m, s = 65390 1/m.
Consequently, two major parameters for the porous-jump in
783
scholar.waset.org/1999.7/15651
World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
FLUENT, the face permeability α and the pressure jump
coefficient C, were defined and determined as follows.
α = μδ/A = ε2/(aQs2) = 2.634438x10-12 m2
(3)
C = 2B/ρδ = 2bQ/(ε2D) = 8.454735x105 1/m
(4)
Seven cases, each with a different working flow rate, were
examined for sub-cooled LOX at a condition of 1620269 Pa
(235 psi), 90 K to assess the LAD channel performance and
reveal pressure and velocity distribution within the channel.
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
III. CFD SIMULATION RESULTS
The CFD simulation results show that although both
pressure drops across the screen and along the channel change
with the working flow rates, the flow and pressure distribution
patterns are the same for all cases. Typical flow pattern in the
channel is sho wn in Fig. 3, while Fig 4 sho ws the
a)
Pressure distribution in the LAD channel.
Right end of screen
Vortex
b)
a)
b)
Fig.4 Pressure distribution in the LAD channel for the case of
working flow rate 1.096 kg/s.
Velocity vectors distribution in the LAD channel.
Vortex
Magnified view of velocity vectors near the outlet
tube at center section (z=0).
Fig 3 Velocity vectors in the LAD channel for the case of working
flow rate 1.096 kg/s.
International Scholarly and Scientific Research & Innovation 3(10) 2009
Magnified view of pressure distribution near the outlet tube
at center section (z=0).
corresponding pressure distribution pattern.
It can be seen that the LOX passing through the screen flows
towards the outlet-tube. As the flow approaches the outlettube, the velocity increases due to the additional LOX flow
entering the LAD channel. Two vortex zones occur in the
vicinity of the outlet-tube entrance: The first zone is located at
the right end of the channel while the other is located at the
inner side of the outlet-tube under the entrance, as indicated in
Figs. 3b and 4b. These vortices create local low pressure zones
in these areas. Worthy of note is that the pressure drop across
the screen increases along channel length because of the
augmentation of flow rate along the channel length, resulting
in larger flow resistance. As mentioned above, the pressure
distribution patterns remain the same for all of the different
working flow rates. Two typical pressure distributions along
the channel length at different levels, just above the screen (y
= 2 mm), just under the screen (y = - 3mm), and at the channel
center line (y = -12.7 mm), are shown in Fig. 5. For the case of
working flow rate G = 0.237 kg/s, the pressure drop across the
784
scholar.waset.org/1999.7/15651
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
screen at the left end of the screen, far from the outlet-tube (x
= - 241 mm), is Δpsl = 90 Pa, while at thee right end of the
screen, close to the outlet-tube (x = 241 m
mm), the pressure
drop across the screen Δpsr = 127 Pa. For G = 1.488 kg/s, Δpsl
= 510 and Δpsr = 2000 Pa.
In Fig. 5, the area between the vertiical dotted lines
corresponds to the channel length covered byy the screen. The
area beyond these dotted lines corresponds tto the closed area
of the channel. In the right closed area of thhe channel, there
are no pressure losses from an increased LOX flow, and
consequently, the pressure presents a shortt constant region
followed by a sharp decrease because of thhe vortex. This is
depicted in Fig. 5 with the annotation ‘Jusst under screen’.
Obviously, the vortex affects the pressure at tthe channel center
more; therefore, no constant pressure regioon occurs at the
channel center line. The vortex creates a low
w pressure zone in
the channel that yields the largest pressure ddrop, Δptop, at x =
282 mm along the channel center line. For G = 0.237
kg/s, Δptop = 147 Pa, while for of G = 1.488 kkg/s, Δptop = 3250
Pa.
In fact, the lowest pressure in the entirre LAD channel
assembly occurs within the vortex in the outlet-tube. The
step-change in the flow area at the entrance of the outlet-tube
the vena-contracta effect to create a negative pressure zone. A
p (Pa)
Just above screen (y = 2 mm)
Just under screen
(y = - 3 mm)
At channel center liine
(y = - 12.7 mm)
x (m)
ure loss that combines with
produces a considerable large pressu
typical pressure distribution curve in
i the LAD channel along
the outlet-tube center line (y-axis dirrection) is shown in Fig. 6.
For G =1.488 kg/s, the lowest pressure, -500 Pa, occurs at y =
- 40 mm on the outlet-tube center line. That means the
maximum pressure drop in the entiire LAD channel assembly
is 13500 Pa, much greater than thee pressure drop of 3250 Pa
within the channel alone.
I. DISCUSSIIONS
A. Flow Rate across Screen
The CFD simulations revealed th
hat as LOX passes through
the screen and flows toward the outlet-tube in the channel, the
flow velocity and rate in the channeel is continuously augment
along the channel length, x-axis direction, as shown in Fig. 7.
ntation along the channel
This continuous flow rate augmen
increases the flow resistance resultin
ng in a continuous pressure
drop along the channel. Consequ
uently, the pressure drop
across the screen increases along th
he channel length, resulting
in a corresponding increase along the channel length in the
n.
flow rate passing through the screen
Figure 8 shows the distributions of velocity flowing out of
the screen for the cases of G = 1.096 kg/s and G = 0.237 kg/s.
w profiles are different for
It can be seen that although the flow
the different working flow rates, th
he general tendency of the
local velocity passing through the screen is incremental along
the channel length. In the left portiion of the screen, far from
the outlet tube, lower velocities occcur along the edges and in
the central region, while closer to th
he outlet-tube, the velocity
in the middle reaches maximum and gradually decreases
toward the edges.
Obviously, the position of the outtlet-tube greatly affects the
flow pattern of LOX passing through
h the screen. As it is seen in
Figs. 1 and 4b, the outlet-tube is loccated at channel’s right end
and creates a very low pressure zo
one in the channel by the
entrance of the outlet-tube, as seen in Figs. 4 and 5.
Consequently, the local flow rate off LOX passing through the
screen is significantly larger at the screen’s right end,
a) For G = 0.237 kg/s.
Just above screen (y = 2 mm)
p (Pa)
p (Pa)
Just under screen
(y = - 3 mm)
At channel center line
(y = - 12.7 mm)
y (m)
x (m)
b) For G = 1.488 kg/s.
Fig 5. Pressure distributions in the LAD channel at different levels.
International Scholarly and Scientific Research & Innovation 3(10) 2009
Fig. 6 Pressure distribution along the outtlet-tube center length (y-axis
direction) for the case of G = 1.488 kg/s.
785
scholar.waset.org/1999.7/15651
World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
especially along the screen centerline, as shown in Figs. 7 and
8.
B. Flow Resistance in the LAD Channel
Since the flow in the channel is continuously merging with
flow from across the screen the flow is subjected to an
increasing flow resistance along the channel. Obviously, this
flow resistance is neither the simple frictional resistance nor the
geometrical resistance created by change in the cross-sectional
flow area. Both of these are usually ignored for their relatively
small contributions. Previous investigators may have
considered this as a frictional resistance. The flow resistance
Flow out of screen at middle
of the rightmost end
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
a)
a)
For the case of G = 1.096 kg/s
At leftmost end of screen (x = -0.24 m)
b)
For the case of G = 0.237kg/s
Fig. 8 Velocity distribution on the screen surface facing to the
channel for different cases of flow rates
b)
At middle of screen (x = 0.0 m)
Flow out of
screen
c)
At rightmost end of screen (x = 0.24 m)
Fig. 7 Velocity vectors at different y-z sections for the case of G =
1.096 kg/s.
International Scholarly and Scientific Research & Innovation 3(10) 2009
produced by the impact of the flow entering from across the
screen and the mergence with the flow in the channel, Δpim, is
relatively small for slow flow rates, but grows very rapidly
with the working flow rate. For example, from Fig. 5, for G =
0.237 kg/s, Δpim = Δpsr - Δpsl = 37 Pa which is only 0.41times
the pressure drop across the screen, i.e. Δpim = 0.41 Δpsl.
However, for G = 1.488 kg/s, Δpim = Δpsr - Δpsl = 1490 Pa or
2.92 Δpsl. The mathematical expression of the relationship
between the resistance Δpim and the working flow rate G has
not been developed. More experimental and theory studies are
needed.
C. Bubble Point Limitation
As it is well known, the total pressure loss in a LAD
channel must be less than the bubble point pressure to prevent
vapor ingestion into the LAD channel. According to the
measurement results by Jurns and McQuillen [5], for the
screen of Dutch Twill 200 x 1400 mesh, the average value of
bubble points of LOX at 161.7 R (89.83 K) is 2500.8 Pa.
786
scholar.waset.org/1999.7/15651
International Science Index, Physical and Mathematical Sciences Vol:3, No:10, 2009 waset.org/Publication/15651
World Academy of Science, Engineering and Technology
International Journal of Mathematical, Computational, Physical, Electrical and Computer Engineering Vol:3, No:10, 2009
However, the experimental data ranged between 2115.1 to
2737.2 Pa. The limitation of the total pressure loss in the
LAD channel should be put at the low end of the experimental
data for the safe consideration, i.e. the limiting total pressure
loss should be 2115 Pa.
Although this limitation is that the total pressure loss in the
system must be less than the bubble point pressure, the
pressure losses in the outlet-tube should be ruled out if the
outlet-tube does not contribute towards the liquid acquisition
through a screen-type construction. However, the pressure loss
within the screen channel should be included. Three strategic
pressure drops in the channel vs. the working flow rate are
plotted in Fig. 9.
It can be seen that for the test LAD channel simulated here,
when the working flow rate G > 1.22 kg/s (2.69 lb/s), the
largest pressure drop in the channel would exceed the bubble
point pressure. Thus, the performance of this LAD channel
assembly is acceptable for flow rates less 1.22 kg/s. However,
if the vortices at the outlet-tube can be eliminated by changing
the position and alignment of the outlet-tube, the working flow
rate limitation would increased to at least 1.577 kg/s. The
simplest construction of the LAD channel assembly is to align
the outlet-tube with the LAD channel length, as was in [2].
Furthermore, a gradual, smooth connection between the
channel and the outlet-tube, such as a horn-type connection,
can further reduce the size of the vortex zone. This may also
improve the flow rate pattern of the flow passing through the
screen. A more even flow rate distribution over the screen
would decrease the flow resistance created by the flows
impact and mergence. It should be pointed out, however, that
an aligned outlet-tube and LAD channel may be impractical in
terms of integration inside an actual propellant tank.
Δp (Pa)
3000
Bubble point limitation
2000
Δptop
1000
Δpsl
0.0
1.0
1.5
2.0
G (kg/s)
around the outlet-tube entrance can be eliminated or reduced
by changing the position and alignment of the outlet-tube, the
working flow rate limitation would be increased to at least
1.577 kg/s. A combination of using an aligned outlet-tube and
a horn-type connection between the channel and the outlettube is one method to improve the LAD channel assembly
performance.
REFERENCES
[1]
II. CONCLUSIONS
International Scholarly and Scientific Research & Innovation 3(10) 2009
0.5
Fig. 9 Three strategic pressure drops in the channel vs. flow rate
[2]
The CFD simulation revealed that the flow passing through
the screen is continuously augmented along the channel length
rather than a uniform flow distribution along the channel
length. The outlet-tube position and alignment strongly affect
both flow passing through the screen and the flow in the
channel. In the configuration that was analyzed, two vortices
were created around the entrance of the outlet-tube and
produced two low pressure zones that greatly increased the
total pressure loss in the assembly.
Because the flow rate in the channel is augmented along the
channel length, the flow is subjected to increasing resistance
along this length. This flow resistance is neither simple
friction resistance nor the resistance created by the change in
flow area, it is produced by the flow entering from across the
screen impacting and merging with the channel flow. The
mathematical expression of the relationship between the
resistance Δpim and the working flow rate G has not been
developed. More experimental and theory studies are needed.
In order to avoid bubble breakthrough, i.e, the bubble point
pressure, the flow rate should not exceed 1.22 kg/s for the
current test LAD channel assembly. However, if the vortices
Δpsr
[3]
[4]
[5]
[6]
[7]
787
D. A. Fester, A. J. Villars, and P. E. Uney, “Surface tension propellant
acquisition system technology for space shuttle reaction control tanks,”
J. Spacecraft, vol. 13, No.9, pp. 522-527, 1976,.
M. T. Kudlac, J. M. Jurns, “Screen channel liquid acquisition devices for
liquid oxygen,” presented at the 42nd AIAA/ASME/SAE/ASEE Joint
Propulsion Conference & Exhibit, Sacramento, CA., July 9-12, 2006,
Paper AIAA 2006-5054.
M. V. Dyke, “Identification of influential factors for liquid acquisition
device designs,” presented at 34th Joint Propulsion Conference and
Exhibit, Cleveland, OH, July 13-15, 1998, Paper AIAA-98-3198.
D. J. Chato, and M. T. Kudlac, “Screen channel liquid acquisition
devices
for
cryogenic
propellants,”
presented
at
38th
AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit,
Indianapolis, IN., July 7-10, 2002, Paper AIAA 2002-3983.
J. M. Jurns, and J. B. McQuillen, “Liquid acquisition device testing with
sub-cooled liquid oxygen,” presented at 44th AIAA/ASME/SAE/ASEE
Joint Propulsion Conference & Exhibit, Hartford, CT., July 21-23, 2008,
Paper AIAA 2008-4943.
J. C. Armour, and J. N. Cannon, “Fluid flow through woven screens,”
Journal AIChE, vol. 14, No. 3, pp. 415-420, 1968.
E. C. Cady, “Study of thermodynamic vent and screen baffle integration
for orbital storage and transfer of liquid hydrogen,” NASA-cr-134482,
August 1973.
scholar.waset.org/1999.7/15651